On 11/7/07, Mauro Talevi <[EMAIL PROTECTED]> wrote: > > Some issues to ponder on for my part are: > > - what is the state of IDE integration - really difficult to go back to > state of little or no refactoring support.
Git is a non-starter because it doesn't run on windows and only just works on OSX (it's a bunch of shell scripts). A purely non-scientific poll put mercurial ahead of darcs, because a couple of people I trust prefer it (Joe being one of them) and because it's written in python so more people are hacking extensions for it and it seems to have a livelier community. As for IDE integration, there's surprisingly little you need. You can do some work, then run hg addremove which basically says "I've made some changes you didn't know about - go figure what they are. There's a really nice switch --similarity nn which says that a removed file and an added file count as a rename if they are nn% similar (which they will be if they just have a class rename or a different package). It's surprisingly good at catching stuff. - should we do a comparison of hg vs git? I've never tried or looked at > hg before, while I have done of bit of playing around with git. So, no > real experience to offer, just a let's consider what plays best. IMO, > one key factor in the choice would be the interop with existing more > mature SCMs, eg SVN, which have better IDE support. See above. Mercurial has some nice subversion integrations, and there's even a magic extension that migrates between subversion and mercurial repository histories (as much as it can given mercurial's non-linear model). Food for thought. Indeed. I've found mercurial (and darcs) to be the nicest technologies I've come across in some time. They seem to Just Work in a way that's all too rare. In fact, I'm using mercurial on a client project to manage a three-way merge that subversion is barfing on (which is annoying since it's subversion's three-way merge in the first place!) In fact, I've been meaning to blog my experiences with mercurial on a subversion repository - it's a good combination of offline versioning and the wider acceptance of svn. If I were a betting man I would say to look out for a big shift towards distributed scm on open source projects over the next 18 months. Cheers Cheers, Dan
